US20150227609A1 - Automatic group formation and group detection through media recognition - Google Patents

Abstract

Disclosed is a method and system for automatically forming a group to share a media file or adding the media file to an existing group. The method and system includes receiving, by a processor, a media file comprising an image, identifying, by the processor, the face of a first person in the image and the face of a second person in the image, identifying, by the processor, a first profile of the first person from the face of the first person and a second profile of the second person from the face of the second person, automatically forming, by the processor, a group of recipients in a messaging application, the forming of the group of recipients based on the first profile and the second profile, the group of recipients comprising the first person and the second person; and sharing, by the processor, the media file with the group.

Description

CROSS REFERENCE TO RELATED APPLICATIONS
• This patent application claims the benefit of U.S. Provisional Patent Application Ser. No. 61/939,603, titled “Automatic Group Formation and Group Detection Through Media Recognition” filed on Feb. 13, 2014, the contents of which are incorporated herein by reference.
TECHNICAL FIELD
• The subject matter described herein generally relates to the field of media sharing and social networking. Specifically, the present disclosure relates to automatically defining a group of members within a media sharing service based on tags automatically generated during the processing of a media file.
BACKGROUND INFORMATION
• As portable communication devices such as cellular phones, tablet computers, and laptop computers continue to evolve, improved audio, video, and still image capturing features are typically introduced with each new model. At the same time, the opportunity for users of these devices to capture and share various forms of content with others continues to increase.
• In some situations, a user may distribute content in the form of media files to others by way of specifically determining the recipients of the content by manually attaching the media file to an e-mail, “picture mail,” or similar communication message, and then transmitting the message and attached content from the user's communication device to one or more recipients. Typically, the user distributing the content has pre-programmed certain contact information, such as an e-mail address or telephone number. Alternatively, the user's communication device may also be connected to a recipient via a social network.
• However, the task of manually attaching media files and selecting the relevant recipient information can get tedious and cumbersome if either the quantity of media files or number of recipients becomes too large. Also, in such a scenario, the user is burdened with the task of manually grouping together media files that may be related. Thus, when a user wants to share a special moment, he or she is forced to spend time sorting, labeling, and distributing media files rather than focusing on the actual event.
SUMMARY OF THE INVENTION
• The present disclosure describes methods and systems for facilitating automated group definitions for sharing content through the utilization of media recognition methods. Two features of such automated group definition include (1) determining the formation of new groups and (2) the detection of existing groups.
• In one aspect, a system and method may use a facial recognition algorithm to detect the presence of one or more faces that indicates the presence of one or more persons in a media file. The system may then use a data sample of a detected individual's face to gain information about the individual by sending queries to a database or social network to identify the person's name and contact information. After identifying the name of an individual in the media file, the system can create a recipient list based off of the name and the individual's corresponding contact information. Similarly, the system and method may use scene and object recognition algorithms to suggest a title for the group or a caption for shared content. Finally, the system and method may use location and network information to suggest additional members of the group.
• In one embodiment, the system and method may detect the existence of a group that already lists all the individuals identified in a media file. If such an existing group is found, the system will contribute the new media files to the existing group. Similarly, scene and object recognition algorithms may detect if there is an existing group with a related title, subject or content and the system may contribute the new media files to that group. Finally, location and network detection methods may detect if there are larger groups that have an existing subset group and the system and method may then suggest members to add to the subset group.
• Additional embodiments may implement various combinations of recognition algorithms to detect scenes, objects, location, and networks in media files to automatically create or contribute content to groups.
BRIEF DESCRIPTION OF THE DRAWINGS
• FIG. 1 depicts a user interface for automatically recognizing faces within a photo captured by a camera on a mobile device in accordance with an embodiment of the present disclosure.
• FIG. 2 depicts a user interface for automatically forming a group of members of a media sharing service based on faces recognized in an image in accordance with an embodiment of the present disclosure.
• FIG. 3 is a flowchart illustrating a process in which a media sharing service determines how to group new media files in accordance with an embodiment of the present disclosure.
• FIG. 4 illustrates how a media sharing service may query a database or social network to retrieve personal profile information in accordance with an embodiment of the present disclosure.
• FIG. 5 illustrates exemplary mobile devices that may serve as client devices in accordance with an embodiment of the present disclosure.
DETAILED DESCRIPTION
• For the purpose of promoting an understanding of the principles of the present disclosure, reference will now be made to the embodiments illustrated in the drawings and specific language will be used to describe the same. It will, nevertheless, be understood that no limitation of the scope of the disclosure is thereby intended; any alterations and further modifications of the described or illustrated embodiments, and any further applications of the principles of the disclosure as illustrated therein are contemplated as would normally occur to one skilled in the art to which the disclosure relates. All limitations of scope should be determined in accordance with and as expressed in the claims.
Definitions
• The term “mobile device” is intended to refer to any generally portable device having network access (e.g. Internet access) and including a camera or video capturing mechanism. Of note, such network access may be available to the mobile device wirelessly and/or through a wired mechanism (such a wired mechanism may include a removable plug or the like, so that portability is maintained).
• The term “application” (such as used in the context of an application associated with a mobile device) is intended to refer to a mechanism that provides the functionality described herein.
• The term “camera” (such as used in the context of a camera included with a mobile device or externally connected) is intended to refer to a mechanism that provides still image capture and/or video image capture.
• The term “internet access” is intended to refer to the ability to directly or indirectly connect (e.g., bi-directionally) with the Internet. Internet access may be selected from the group including (but not limited to): access via wi-fi, access via bluetooth, access via a cell phone network (e.g., a 2g network, a 3g network, or a 4g network), and access via a wired mechanism (such a wired mechanism may include a removable plug or the like, so that portability is maintained).
• The term “friend” is intended to refer to a member of a social networking service who has indicated a willingness to be identified as such with respect to another person or persons. The term may also refer to a person as noted in a user's contact list, such as in the address book of an email application or service.
• The term “social network” is intended to refer to a website or application that enables users to communicate with each other by posting information, comments, messages, images, videos, etc. A social networking service may be one of, but not limited to, the following exemplar services: Facebook®, Google+®, Twitter®, Instagram®, Foursquare®, LinkedIn® and Flickr®.
• The term “client system” is intended to mean systems that may send queries to and receive data from a visual query server system. The client system may be any computer or other device that is capable of communicating with the visual query server system. Examples include, without limitation, desktop and notebook computers, mainframe computers, server computers, mobile devices such as mobile phones and personal digital assistants, network terminals, and set-top boxes.
• The term “client device” is intended to mean a device that can operate as a client system.FIG. 5 depicts exemplar mobile devices that may operate as client devices. A client device may be, but may not be limited to, any one of the following: amobile phone510, atablet device520, alaptop computer530.
• The term “visual query” refers to an image or video that is submitted as a query to a search engine or search system. Examples of visual queries, without limitations, include photographs, scanned documents and images, videos and drawings.
• The term “search service” is intended to refer to a system that extracts textual tags of a visual query from a client system and searches the database against those tags these tags include but are not limited to user profiles, objects, scenes and other metadata.
Embodiments
• Example methods and systems for the forming of user groups primarily for the sharing of content, including, for example, photographs, audio content, video content, and textual content, are described. In the following description, for purposes of explanation, numerous details are set forth in order to provide a thorough understanding of example embodiments. However, it will be evident to one skilled in the art that the present subject matter may be practiced without these specific details. It will also be evident that the types of software development described herein are not limited to the examples provided and may include other scenarios not specifically discussed.
User Interfaces
• FIG. 1 depicts auser interface100 for automatically recognizingfaces101a,101b,101cwithin aphoto105 captured by a camera on a client device, according to one embodiment. Thecamera icon120 at the bottom of theuser interface100, when tapped, may capture a photo. Thecamera icon130 on the top right, when tapped, may toggle between the back facing camera and the front facing camera on the client device. Once aphoto105 is captured, the client device may begin detecting faces in theimage110. Once a face is recognized,squares101a,101b,101cmay be placed around the faces. Dots may be used to recognize features and placement unique to each face. These placement and measurements may be used to create a unique visual map that may then be used to query social networks (or another database of media) to uniquely identify each of the faces. The interface depicted inFIG. 1 appears within a proprietary mobile application for group sharing and messaging.
• FIG. 2 depicts auser interface200 for automatically forming auser group210 consisting ofgroup members201a,201b,201cof a media sharing service based, at least in part, on faces recognized in animage205, according to one embodiment. Thisinterface200 may appear once thephoto205 is captured or selected and thefaces101a,101b,101care identified. The “new group”title210 may indicate that a new group is being formed. In this example, Austin, Hamon, and Howard are individuals whose faces were detected and have been automatically populated into the “To:”field220. Once the “Done”button230 is pressed anew user group210 is created and theparticipants201a,201b,201cmay now interact with each other by sharing, among other things, comments, more photos, and additional media. The interface depicted inFIG. 2 appears within a proprietary mobile application for group sharing and messaging.
Group Formation and Detection
• In a client device, auser interface module100 may be used to facilitate a user's access to various aspects of sharedcontent105 within auser group210, including, but not limited to, providing explicit input as to which group to join, posting content to the group, and receiving content posted by other members of theuser group210.
• Specifically,FIG. 3 depicts a system and method in which, when new media is detected300, various applications of media andcontext recognition technology311,321,331 may be leveraged for defining and creating new user groups as well as for contributing to existing user groups. Generally,new groups316,326,336 may be formed when the system determines that the analyzed media does not belong in an existinggroup315,325,335. Similarly when the system determines that the analyzed media belongs to an existinggroup315,325,335, the analyzed media may be shared to the existinguser groups317,327,337, respectively.
• Usingfacial recognition methods310, the analyzed faces311 in a media file may be processed at aserver312 in the form of a visual query being sent to asearch service313. If any of the faces are recognized, the matching profiles314 may be used to populate members in anew group316. When a set of people are recognized in aphoto314, and a user group with that exact same member constellation already exists, the existing user group may be detected315 and user may be given theoption317 to have the new media be shared into thegroup350. Alternatively, a user may choose to not add the new media to agroup360.
• Similarly, the system may also utilize scene and objectrecognition methods320 to help in forming and detecting groups. Using such methods, the analyzed scenes andobjects321 in a media file may be used to determine whether thenew media300 belongs in anew group326. If an existing group already features the recognized scene or objects, the existing user group may be detected325 and the user may be given theoption327 to have the new media be shared into thegroup350.
• Additionally, when appropriate, scene and objectrecognition methods320 may be used to create relevant group titles and media captions. In cases when new user groups are formed, the output of the analysis may be used to populate both user group titles and/or the first message posted to the user group. In the cases where a user group containing the scene or object already exists, the output of the analysis may be used to generate appropriate captions that may be shared in connection with the media.
• Furthermore, location and networkrecognition methods330 may be used to automatically populate the members of a user group as well. With these methods, people in the same vicinity or on the same network, may be prompted to join a newly formed user group consisting of other people in that same vicinity ornetwork336. Alternatively, location and network analysis may be leveraged to add new members to an already existinguser group335, prompting the existing user group members to confirm the addition of new members that were detected on the same network or identified in thevicinity337.
• Alternatively, if none of the aforementioned methods (facial311, object/scene321, and local/network331 methods) detect any usable information, the system may prompt the user to manually group340 thenew media file300.
• Finally, in each of the aforementioned scenarios, the client system and the server stay in agreement with each other as to the particular groups that the client system is a member of. Furthermore, the evaluation process of operations in regards to deciding whether the client system should remain in a particular group (or be removed from the group, possibly in order to join a new group) may be performed again in response to one or more events. Grouping operations may also occur in response to the creation or capture of new content on a client device, which is then uploaded to the server for distribution to other member client systems of the group.
Facial Recognition
• In the present disclosure, the term “facial recognition” refers to a method of processing a visual query of a facial image performed on a client system having one or more processors and memory storing one or more programs. A visual query of media files comprising one or more facial images is mapped, and possible images that may match with respective facial images are identified in accordance with visual similarity criteria. The potential image matches comprise images from one or more image sources identified in accordance with data regarding the requester. The image match results in one or more persons associated with the potential image matches being identified.
• The visual query server system includes a front-end visual query-processing server. The front-end server receives a visual query from the client, and sends the visual query to a plurality of parallel search systems for simultaneous processing. A face recognition search system will access a facial image database to look for facial matches to the image query.
• In one embodiment, a media sharing system interacts with a social networking service to identify recognized faces. In such an embodiment, the social networking service may have associated member profiles corresponding to one or more respective members, wherein each of the member profiles includes at least one member image associated with a respective member. Furthermore, each member image includes a representation of a face of the respective member.
• In such an embodiment, a client system may send at least one source image, wherein the source image includes a representation of a face. A search service that may be implemented at a server compares at least one of the representations of the faces in the source image to at least some of the representations of the faces in the member images. The search service then communicates to the client system at least one result identifying at least one member who has a member image including a representation of a face that is a match to a representation of a face in the source image. The client system may then use this information to contact that member.
Scene Detection
• Another method for sorting media, in the present disclosure, is by grouping media based on detecting the scene in an image or video. The scene representation may be a recognizable instance of the scene based on a realistic, e.g., photographic, drawn, painted, caricatured, cartoon, or other recognizable representation of the scene. The scene may also be a certain scene based on a predefined type or category of scenes or other recognizable scenes with a distinct appearance that distinguishes them from other scenes.
• The scene detector includes a classifier for example a classification function based on one or more features of a scene that evaluates the image or part of an image to determine if the scene can be detected or not. The scene detector further includes an adaptive image scanner that processes the input image to divide the image into smaller windows.
• The adaptive image scanner scans the input image to divide the image into scanning windows and utilizes a classifier to classify each sub window. The scanning window can be classified with respect to a cascade of homogenous classification functions covering one or more features of the scene. Thereafter, a determination can be made whether a positive detection of the scene can be encountered. If a positive detection is not encountered, scanning may be continued. Otherwise, a reasonable neighborhood can be established such that a higher probability of other detections can be expected. The neighborhood may be searched utilizing fine shift methods in order to discover additional detections. The scene detector outputs an output image with the encountered detections. After the local neighborhood is sought through, the scan can be continued with the rough step corresponding to the actual scale. These encountered detections may be collected and utilized for further processing.
Object Recognition
• Yet another method for sorting media, in the present disclosure, is by grouping media based on detecting an object in an image or video. The object representation can be a recognizable instance of the object based on a realistic representation such as a photographic, drawn, painted, caricatured, cartooned, or other recognizable representation of the object. The object may further be a certain object based on a predefined type or category of objects, such as faces, dogs, cars, trees, or other recognizable objects with a distinct appearance that distinguishes them from other objects.
• An object detector may further include a classifier for example a classification function based on one or more features of an object that evaluates the image or part of an image to determine if the object can be detected or not. The object detector may also include an adaptive image scanner to process the input image to divide the image into smaller windows.
• The adaptive image scanner scans the input image to divide the image into scanning windows and utilizes a classifier to classify each sub window. The scanning window may be classified with respect to a cascade of homogenous classification functions covering one or more features of the object. Next, a determination may be made regarding whether a positive detection of the object may be encountered.
• In the event that a positive detection is not encountered, scanning may proceed. Otherwise, a reasonable neighborhood can be established such that a higher probability of other detections can be expected. The neighborhood may be searched utilizing fine shift methods in order to discover additional detections. The object detector outputs an output image with the encountered detections. After the local neighborhood is searched through, the scan may be continued with the rough step corresponding to the actual scale. Encountered detections may then be collected and utilized for further processing.
Location Detection
• Another embodiment relates to utilizing location data stored in a media file that was captured using a client device. In one example, the media may be a photograph that is stored as data on a computer-readable medium. The location of the user at the time the photograph was captured may thus correspond to the location of the client device at the time the photograph was captured. The particular media on the client device may further include aperture, shutter speed, ISO, camera mode, focal distance and GPS information. Further, such information may be stored in an EXIF data location on the client device.
• In a group detection embodiment, when a person shares photos with a given group of people, and persons in that group have taken photos in the same location, a recommendation is sent to those persons to share photos from the same location into the group. In a new group formation embodiment, when users share photos with the same location and time, a recommendation is sent to form a group. Furthermore, in situations where multiple individuals are located at the same location or on the same network, then such points of relevance may be utilized to form a group. Therefore, other individuals on the network or at a similar physical location do not necessarily have to take a picture themselves to become part of a group, or receive shared media. Furthermore, physical location relevance may be customized in the algorithm to account for how much distance would still be considered a relevant potential group member.
Network Detection
• Network detection is an embodiment where a mobile device may be connected to via Wi-Fi to a local area network and may have the capability of transmitting and receiving data (e.g., voice, text, images, and/or multimedia data). When the same application on the client device is launched, other client systems on the network may be able to detect the presence of other client devices with the same application active.
• Users of a mobile application on the same network may be able to automatically create a group with all other users of the mobile application on the same network. Group detection occurs when users on the network are already in a group together, then a recommendation is sent to share media. Group formation occurs when all users on the network are populated in the “to:” field of a new group compose screen.
Identification
• FIG. 4 depicts interactions between a client system and a server when processing a new media file. In one embodiment, aserver system460 may receive avisual query423 from aclient system450. Theclient system450, for example, may be amobile phone510, atablet device520, adesktop computing device530, or another similar device. When new media is created411, theclient system450 may process thenew media420 for the detection of faces, scenes, and objects421. Additionally, theclient system450 may process the media to extractuseful metadata422. Having performed these tasks, theclient system450 may then form a request to theserver460 in the form of avisual query423 comprising of thumbnail pictures, atlas, and metadata information.
• In an example scenario, in afirst step410, a user may take a photograph of hisfriend411 using a mobile phone user interface412. Subsequently, theclient system450 may submit processed parts of thephotograph421422 as thevisual query423 to theserver system460.
• At theserver system460, thevisual query423 may be processed, via asearch service430, by having the image component of the query compared amongst a database of identified persons, objects, andscenes431. In one embodiment, for each identifiable person in thedatabase431, there exists at least one photograph and a profile containing descriptive information. Such photos and descriptive information may be retrieved from a social networking service. The social networking service may be one of, but not limited to, the following exemplar services: Facebook®, Google+®, Twitter®, Instagram®, Foursquare®, LinkedIn® and Flickr®.
• In one embodiment, theimage processing engine430 and training setdatabase431 operate in conjunction as asearch service435 in order to extract textual tags of avisual query423 from aclient system450 and search thedatabase431 against those tags. Such a search may result in, but not limited to, facial recognition, object recognition, and scene detection in an image. The results of the search may be fed toapplication logic433 and accompanyingdatabase432.Application logic433 and accompanyingdatabase432 may generate suggestions for, but not limited to, groups, captions, and tags. These suggestions are presented to the user in a form of aconfirmation user interface442. Lastly, in asecond confirmation step440, the user either confirms or rejects thesuggestions441. Thisuser feedback450 is then committed to the accompanyingdatabase432. Additionally,user feedback data451 is incorporated into thetraining set database431 to improve training
Training
• With respect to training in certain embodiments, compartmentalization of information is expected. If a user is sharing photos, the ability to recognize particular people is dependent upon the data set the user has access to. In such an example, each person effectively has their own recognition model. In one embodiment, for facial recognition, theserver system460 has positive and negative training examples. For example, theserver system460 may have 10 photos where a user was correctly tagged in the photos and 2 photos where the server incorrectly tagged the user in the photos. These examples can merge into a model that can recognize a face with some confidence. Each user can have their own training set, and only photos that the user can see can become part of the training set. Given that, each user can have their own recognition model based on the photos they have access to. In one embodiment, there is no side channel leakage of information.
• In one embodiment of training, theserver system460 ranks, based on the model, a set of candidates during facial recognition. A recognizer can provide a user with N best matches of results during facial recognition with a confidence score for each of the N best matches. In one embodiment, theserver system460 provides the user with the best match for a face. In another embodiment, theserver system460 provides the user with two or more identities of user profiles for a face and lets the user select which identity is the correct identity for the face. Among a set of different faces, theserver system460 can provide recommendations based on a confidence score associated with the identities.
• Thus, a method and system to define groups, through image recognition, to share content among several communication devices has been described. Although the present subject matter has been described with reference to specific example embodiments, it will be evident that various modifications and changes may be made to these embodiments without departing from the broader scope of the subject matter. For example, while the majority of the discussion above notes the use of the embodiments with respect to computer systems and applications, other software- or firmware-based systems, such as electronic products and systems employing embedded firmware, may also be developed in a similar manner to that discussed herein. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense. The accompanying drawings that form a part hereof show, by way of illustration and not of limitation, specific embodiments in which the subject matter may be practiced. The embodiments illustrated are described in sufficient detail to enable those skilled in the art to practice the teachings disclosed herein. Other embodiments may be utilized and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. This detailed description, therefore, is not to be taken in a limiting sense.
• Thus, although specific embodiments have been illustrated and described herein, it should be appreciated that any arrangement calculated to achieve the same purpose may be substituted for the specific embodiments shown. This disclosure is intended to cover any and all adaptations or variations of various embodiments. Combinations of the above embodiments, and other embodiments not specifically described herein, will be apparent to those of skill in the art upon reviewing the above description.

Claims (22)

What is claimed is:

1. A method comprising:

receiving, by a processor, a visual query comprising data related to an image captured by a client device, the image comprising faces;

identifying, by the processor, by applying a facial recognition algorithm to facial data of people in a facial data database, the face of a first person in the image and the face of a second person in the image;

identifying, by the processor, by analyzing people profiles in a profile database, a first profile of the first person from the face of the first person and a second profile of the second person from the face of the second person;

determining, by the processor, whether an existing group of recipients in a media sharing service exists, the existing group comprising the first person and the second person;

when the existing group exists, transmitting, by the processor, instructions to the client device to associate the image with the existing group;

when the existing group does not exist, automatically causing, by the processor, formation of a new group of recipients, the formation of the new group based on the first profile and the second profile, the new group of recipients comprising the first person and the second person; and

when the new group is formed, transmitting, by the processor, instructions to the client device to associate the image with the new group.

2. The method ofclaim 1, further comprising:

adding, by the processor, the image associated with the identified face of the first person and the identified face of the second person to a training set.

3. The method ofclaim 1, wherein the visual query refers to a submitted image or video submitted as a query to the processor.

4. The method ofclaim 1, wherein the data comprises metadata obtained from the image.

5. The method ofclaim 1, further comprising extracting, by the processor, textual tags of the visual query.

6. The method ofclaim 5, further comprising searching the profile database against those tags.

7. The method ofclaim 1, wherein the profile database comprises a database of a social networking service.

8. A method comprising:

receiving, by a processor, data related to an image captured by a client device, the image comprising a recognizable element selected from a group of recognizable elements consisting of a scene, an object, a location, and a network;

identifying, by the processor, the recognizable element by applying a recognition algorithm to recognizable element data in a recognizable element data database;

determining, by the processor, whether an existing group in a media sharing service exists, the existing group comprising the recognizable element;

when the existing group with the recognizable element exists, transmitting, by the processor, instructions to the client device to associate the image with the existing group;

when the existing group with the recognizable element does not exist, automatically causing, by the processor, formation of a new group, the formation of the new group based on the recognizable element; and

when the new group is formed, transmitting, by the processor, instructions to the client device to associate the image with the new group.

9. The method ofclaim 8, wherein when the recognizable element is a scene or object, creating, by the processor, a title or a caption associated with the existing group or the new group.

10. The method ofclaim 8, further comprising receiving, by the processor from the client device, an approval to have the image added to the existing group.

11. The method ofclaim 8, wherein when the recognizable element is a location, determining, by the processor, whether the client device is in substantially a same location as other members of the existing group or new group.

12. The method ofclaim 8, wherein when the recognizable element is a network, determining, by the processor, whether the client device is on a same local network as other members of the existing group or new group.

13. A method comprising:

storing, by a processor, positive and negative training examples, the training examples comprising a first set of examples where a user was correctly tagged in a first set of images and a second set of examples where the user was incorrectly tagged in a second set of images;

merging, by the processor, the positive and negative training examples into a model for the user, the model facilitating recognizing faces in a plurality of images;

receiving, by the processor, a visual query comprising data related to a captured image captured by a client device, the captured image comprising faces;

ranking, by the processor and based on the model, a set of candidates that potentially match the face of a first person in the captured image and the face of a second person in the captured image;

based on the ranking, identifying, by the processor, a best match from the set of candidates for the face of a first person and a best match from the set of candidates for the face of the second person; and

identifying, by the processor, by analyzing people profiles in a profile database, a first profile of the first person from the best match for the face of the first person and a second profile of the second person from the best match for the face of the second person.

14. A method comprising:

storing, by a processor, positive and negative training examples, the training examples comprising a first set of examples where a user was correctly tagged in a first set of images and a second set of examples where the user was incorrectly tagged in a second set of images;

merging, by the processor, the positive and negative training examples into a model for the user, the model facilitating recognizing faces in a plurality of images;

receiving, by the processor, a visual query comprising data related to a captured image captured by a client device, the captured image comprising faces;

ranking, by the processor and based on the model, a first set of candidates that potentially match the face of a first person in the captured image and a second set of candidates that potentially match the face of a second person in the captured image;

transmitting, by the processor to the client device, the first set of candidates;

receiving, by the processor from the client device, a selection of one of the first set of candidates as a correct match for the face of the first person in the captured image;

transmitting, by the processor to the client device, the second set of candidates;

receiving, by the processor from the client device, a selection of one of the second set of candidates as a correct match for the face of the second person in the captured image;

identifying, by the processor, by analyzing people profiles in a profile database, a first profile of the first person from the selection of one of the first set of candidates and a second profile of the second person from the selection of one of the second set of candidates;

incorporating, by the processor, the selection of one of the first set of candidates into the model; and

incorporating, by the processor, the selection of one of the second set of candidates into the model.

15. The method ofclaim 14, wherein the incorporating of the selection of one of the first set of candidates into the model further comprises storing the selection in a database.

16. The method ofclaim 14, wherein the incorporating of the selection of one of the second set of candidates into the model further comprises storing the selection in a database.

17. A computing device comprising:

a processor;

a storage medium for tangibly storing thereon program logic for execution by the processor, the stored program logic comprising:

visual query receiving logic executed by the processor for receiving a visual query comprising data related to an image captured by a client device, the image comprising faces;

face identifying logic executed by the processor for identifying, by applying a facial recognition algorithm to facial data of people in a facial data database, the face of a first person in the image and the face of a second person in the image;

profile identifying logic executed by the processor for identifying, by analyzing people profiles in a profile database, a first profile of the first person from the face of the first person and a second profile of the second person from the face of the second person;

existing group determining logic executed by the processor for determining whether an existing group of recipients in a media sharing service exists, the existing group comprising the first person and the second person;

existing group image adding logic executed by the processor for, when the existing group exists, transmitting instructions to the client device to associate the image with the existing group;

new group formation logic executed by the processor for, when the existing group does not exist, automatically causing formation of a new group of recipients, the formation of the new group based on the first profile and the second profile, the new group of recipients comprising the first person and the second person; and

new group associating image logic executed by the processor for, when the new group is formed, transmitting instructions to the client device to associate the image with the new group.

18. The computing device ofclaim 17, further comprising:

training set adding logic executed by the processor for adding the image associated with the identified face of the first person and the identified face of the second person to a training set.

19. The computing device ofclaim 17, wherein the visual query refers to a submitted image or video submitted as a query to the processor.

20. The computing device ofclaim 17, wherein the profile database comprises a database of a social networking service.

21. A computing device comprising:

data receiving logic executed by the processor for receiving data related to an image captured by a client device, the image comprising a recognizable element selected from a group of recognizable elements consisting of a scene, an object, a location, and a network;

element identifying logic executed by the processor for identifying the recognizable element by applying a recognition algorithm to recognizable element data in a recognizable element data database;

existing group determining logic executed by the processor for determining whether an existing group in a media sharing service exists, the existing group comprising the recognizable element;

existing group image adding logic executed by the processor for, when the existing group exists, transmitting instructions to the client device to associate the image with the existing group;

new group formation logic executed by the processor for, when the existing group does not exist, automatically causing formation of a new group, the formation of the new group based on the recognizable element; and

new group associating image logic executed by the processor for, when the new group is formed, transmitting instructions to the client device to associate the image with the new group.

22. The computing device ofclaim 19, further comprising location determining logic executed by the processor for, when the recognizable element is a location, determining whether the client device is in substantially a same location as other members of the existing group or new group.

Images